Representations of Knowledge & Language–Cognition Study Notes

UNIT 1 – MENTAL REPRESENTATION OF KNOWLEDGE

• Topic Learning Outcome: discuss how knowledge is stored/represented in mind.

ENGAGE

• Quick exercise: students produce own mental representations (pictures, words, propositions).

EXPLORE – What is Mental Representation?

• Core question: How is knowledge stored in mind/brain?
• Forms of internal code:
– mental pictures (analog, modality-specific)
– verbal strings (symbolic, sequential)
– abstract propositions (amodal, logic-like)
• Empirical approaches:
a) standard laboratory experiments
b) neuro-psychological studies (lesion, imagery deficit, imaging).

EXPLAIN – Pictures vs Words

• Both act as representational surrogates for external reality, yet differ fundamentally.
– Pictures: concrete, depictive, spatially simultaneous, relatively analogous; retain info even when partially degraded (dog face w/out body still dog).
– Words: abstract, arbitrary, rule-bound, sequential; any loss or permutation of letters ("dog"→"do" or "god") destroys original reference.
• Neither medium captures all attributes of referent (picture of puppy doesn’t bark; word puppy doesn’t wag tail).

MENTAL IMAGERY

• Definition: internally generated sensory-like experience in absence of external stimulus.
• Modalities: most salient visually, but can involve smell, taste, sound, kinesthesia.
• Imagery can combine:
– memories ("favorite vacation beach")
– novel constructions ("life on another planet")
– impossibilities ("purple sun"), thus underlies imagination/creativity.
• Applied usage: guided imagery in health (pain control, immune boosting); CBT for phobias & anxiety.

DUAL-CODE THEORY (Allan Paivio $1969$, $1971$)

• Mind stores information in two partially independent yet interacting codes:

1. pictorial (image-based)

2. verbal (logogen-based).
   • Encoding advantage: redundancy ⇒ higher memorability (pictures + words > either alone).
   • Empirical finding:
   – Free-order recall superior for pictures.
   – Serial-order recall superior for words.

PROPOSITIONAL THEORY (Anderson & Bower $1973$; Pylyshyn $1973$, $1984$, $2006$)

• Rejects depictive & verbal storage; instead postulates a format of ‘mentalese’.
• Proposition = minimal, language-free unit of meaning expressing a relation.
Syntax of predicate calculus:
$(Relation)(Subject,Object)$
e.g., $(on ext{‐top})(pug,swing)$; $(lifts)(swing,pug)$.
• Relation types: actions, attributes, spatial positions, class membership.
• Complex scenes built by conjoining propositions ("tiny pug… fell… barked… resting under swing").
• Retrieval: deep propositional code reconstructed into either words or images as task demands.

PRACTICE – Predicate Calculus

• Students pick photographs; translate visible relations into predicate statements to appreciate abstraction.

UNIT 2 – DECLARATIVE & PROCEDURAL KNOWLEDGE

• Learning Outcome: Elaborate two knowledge structures.

DEFINITIONS

• Declarative ("knowing what/that"): explicit facts/events.
• Procedural ("knowing how"): algorithms, skills, conditioned associations.

ENGAGE

• Alphabet task: write $A\rightarrow Z$ quickly (routine forward = procedural + declarative).
• Then write $Z\rightarrow A$ (requires declarative only: slower, less fluent). Illustrates synergy vs isolation.

ORGANIZING DECLARATIVE KNOWLEDGE

A. Concepts & Categories

• Concept = mental symbol/unifying idea.
• Category = set of entities grouped by shared criteria or resemblance to prototype.
• Types:
– Natural ("birds", "trees").
– Artifacts ("cars", "toasters").
– Ad hoc (contextual, e.g., "soft things you can hug").

B. Distinctions in Category Structure

• Classical concepts: sharply defined via necessary & sufficient features ("bachelor = male + adult + unmarried").
• Fuzzy concepts: graded, prototype-based ("game", "death").

C. Approaches to Categorization

1. Feature-Based (defining features).

2. Prototype Theory (characteristic features; typicality effects).

3. Exemplar Approach (multiple stored examples).

4. Core + Prototype Hybrid (core = definitional; prototype = typical).

5. Theory-Based / Explanation-Based (implicit naive theories; e.g., concept of "wise person" involves causal-explanatory beliefs).

D. Semantic Network Models

• Knowledge stored as nodes (concepts) linked by relations (is-a, has-part, can-do).
• Collins & Quillian hierarchical network: economy via cognitive economy (properties stored at highest applicable node).
– E.g., "can fly" stored at "bird", not individually at "robin", "canary".
• Account for typicality & sentence-verification times (distance = processing time).

E. Schemas & Scripts

• Schema = task-oriented mental framework aggregating relevant concepts ("living room" schema includes sofa, TV, etc.).
Pros: guide perception, fill gaps; Cons: stereotype risk.
• Script = schema specifying stereotyped temporal order.
Doctor-clinic example: actors, props, scenes, outcomes.
Scripts supply default values & expectations; aid comprehension, memory, inference.

PROCEDURAL KNOWLEDGE

• Acquisition via practice ⇒ proceduralization (Anderson’s ACT theory).
• Largely implicit once compiled; difficult to verbalize (driving, bicycle).
• Broader non-declarative umbrella:
– perceptual-motor-cognitive skills
– classical & operant conditioning
– habituation/sensitization
– priming (network activation spreads; e.g., ambiguous symbol $13$ perceived as number or letter when primed).
• Retrieval faster than explicit fact recall.

MODULE 6 – LANGUAGE & COGNITION

Fascinating Numerical Facts (all figures in LaTeX)

• Languages worldwide ≈ $(7000)$.
• English dictionary ≈ $(171\,000)$ entries; Chinese most-spoken.
• Khmer alphabet length $=74$ characters; Rotokas $=12$.
• Bible full translations $=554$; Pinocchio $=250$.
• Top linguistic diversity: Papua New Guinea >800; Indonesia >700.
• Most common Latin letter ‘E’ ≈ $11\%$ of Oxford corpus.

Language vs Communication

• Language defined by:
– Regularity (grammar rules).
– Productivity (infinite novel utterances).
– Arbitrariness (form–meaning link not necessary).
– Discreteness (units combinable).
• Bee dances & bird songs = communication but fail arbitrariness/productivity ⇒ not languages.

Language Influences Thought – Historical Perspective

• Wilhelm von Humboldt: language = thought (Weltanschauung hypothesis).
• Sapir-Whorf (linguistic determinism/relativity): language filters reality.
– Strong form: determinism (cannot think outside linguistic categories).
– Weak form: relativity (language influences but doesn’t imprison thought).

Classical Evidence Examined

• Eskimo snow lexicon; Dani color terms (only "mili" vs "mola").
• Rosch/Heider color memory experiments: Dani & English both show focal-color advantage → challenges strong Whorfian claim.

Contemporary Findings – Bilingual & Cross-Linguistic Studies

• Spatial orientation (Kuuk Thaayorre vs English): use of cardinal directions; superior absolute navigation ⇒ language scaffolds attention.
• English (aspect-focused) vs German (endpoint-focused) event framing (Athanasopoulos $2010$):
– Monolingual Germans match ambiguous video with goal-oriented $40\%$ vs English $25\%$.
– Bilinguals’ matching pattern shifts with language context (verbal load task), showing dynamic cognitive framing.
• Implication: language offers lenses; bilingual minds can switch lenses rapidly.

ETHICAL, PRACTICAL & PHILOSOPHICAL IMPLICATIONS

• Instructional design: leverage dual-code (pair graphics with text).
• Therapeutic imagery: adjunct for pain management & anxiety.
• Cultural sensitivity: schemas/scripts risk stereotyping; educators should encourage flexible categorization.
• Language policy: preserving linguistic diversity maintains alternative cognitive frameworks.

CONNECTIONS TO PREVIOUS LECTURES

• Mental representation principles underpin memory models (working memory encodes via phonological loop & visuospatial sketchpad).
• Categorization research dovetails with reasoning/judgment heuristics (availability, representativeness).
• Proceduralization parallels skill acquisition stages (cognitive → associative → autonomous).

SUMMARY CHEAT-SHEET

• Storage codes: Picture, Word, Proposition.
• Dual-Code = two explicit codes; Propositional = one abstract code.
• Knowledge types: Declarative (facts) vs Procedural (skills).
• Concept organization: defining features, prototypes, exemplars, theory-based.
• Network forms: semantic networks, schemas, scripts.
• Language criteria: regular, productive, arbitrary, discrete.
• Whorf: language shapes thought (weak > strong).
• Bilingual studies: cognitive framing is flexible & context-dependent.